NHE3 Kinase A Regulatory Protein E3KARP Binds the Epithelial Brush Border Na+/H+ Exchanger NHE3 and the Cytoskeletal Protein Ezrin*

Cyclic AMP is a major second messenger that inhibits the brush border Na+/H+exchanger NHE3. We have previously shown that either of two related regulatory proteins, E3KARP or NHERF, is necessary for the cAMP-dependent inhibition of NHE3. In the present study, we characterized the interaction between NHE3 and E3KARP using in vitro binding assays. We found that NHE3 directly binds to E3KARP and that the entirety of the second PSD-95/Dlg/ZO-1 (PDZ) domain plus the carboxyl-terminal domain of E3KARP are required to bind NHE3. E3KARP binds an internal region within the NHE3 C-terminal cytoplasmic tail, defining a new mode of PDZ domain interaction. Analyses of cellular distribution of NHE3 and E3KARP expressed in PS120 fibroblasts show that NHE3 and E3KARP are co-localized on the plasma membrane, but not in a distinct juxtanuclear compartment in which NHE3 is predominantly expressed. The distributions of NHE3 and E3KARP were not affected by treatment with 8-bromo-cAMP. As shown earlier for the human homolog of NHERF, we also found that the cytoskeletal protein ezrin binds to the carboxyl-terminal domain of E3KARP. These results are consistent with the possibility that E3KARP and NHERF may function as scaffold proteins that bind to both NHE3 and ezrin. Since ezrin is a protein kinase A anchoring protein, we suggest that the scaffolding function of E3KARP binding to both ezrin and NHE3 localizes cAMP-dependent protein kinase in the vicinity of the cytoplasmic domain of NHE3, which is phosphorylated by elevated cAMP.

the cAMP-dependent inhibition of NHE3 activity. The first involves an increase in NHE3 phosphorylation level by PKA, 1 and the second requires the presence of a regulatory factor. NHE3 phosphorylation by PKA was initially demonstrated by Moe et al. (3), who showed that in AP-1 cells, a Chinese hamster ovary cell line, the cAMP-elicited inhibition of NHE3 activity was accompanied by an increase in NHE3 phosphorylation level. More recently, Kurashima et al. (4) identified the sites of phosphorylation in NHE3 by PKA in AP-1 cells. In contrast, we previously reconstituted the cAMP-dependent inhibition of NHE3 by transfecting one of two regulatory proteins, E3KARP (NHE3 kinase A regulatory protein 2 ; also named TKA-1) or NHERF (NHE regulatory factor), in the PS120 fibroblast cell line, which lacks these regulatory proteins, and consequently, cAMP does not affect NHE3 activity (5)(6)(7). This demonstrates the necessity of a regulatory protein in cAMP-dependent inhibition of NHE3. E3KARP was cloned from a human lung library by yeast two-hybrid screening using the cytoplasmic domain of NHE3 as bait (5). NHERF was identified by limited trypsinization and cellular fractionation of rabbit renal brush border vesicles (6,8). E3KARP and NHERF are closely related proteins of 337 and 358 aa, respectively. These proteins share 52% identity, and both contain two tandem protein interaction module PDZ domains (5). How these regulatory proteins mediate the inhibitory effect of cAMP is not clear.
Protein-protein interactions are intrinsic to virtually every cellular process. Many of these interactions are mediated by modular domains such as the Src homology SH2 and SH3, pleckstrin homology, and PDZ domains (9 -12). We have shown previously that NHE3 and E3KARP directly interact with each other (5). In this present report, we analyzed the interaction between NHE3 and E3KARP to determine minimal domains necessary for the interaction. We found that the interaction of NHE3 with E3KARP represents a new mode of PDZ binding and that it requires a PDZ domain and a non-PDZ domain of E3KARP. We also found that E3KARP binds to an internal domain within the NHE3 cytoplasmic tail, which has previously been shown to be essential for PKA-dependent inhibition of NHE3 (13). In addition, we show that E3KARP binds to the cytoskeletal protein ezrin, which was previously shown to bind to PKA. These findings suggest a potential model for E3KARP in the inhibition of NHE3 activity by PKA.
Plasmid Constructs-Various domains of E3KARP were generated by PCR and were expressed as recombinant proteins in Escherichia coli. These domains include the first PDZ domain (P1) and the second PDZ domain (P2), expressed individually (P1 or P2) or together as a unit (P1-2); the C-terminal domain (C); and the second PDZ domain and the C-terminal domain as a single unit (P2-C) (Fig. 1). We also expressed clone 42 (C42), which extends from aa 130 to 314 of E3KARP and was one of the partial clones initially obtained from yeast two-hybrid screening (5). Fidelity of the PCR products was confirmed by nucleotide sequencing. The PCR products were cloned into pET30a (Novagen), expressed as hexahistidine (His 6 )-tagged fusion proteins in E. coli, and affinity-purified with Ni 2ϩ -nitrilotriacetic acid (NTA) resin as suggested by the manufacturer (QIAGEN Inc.). The above constructs were also expressed as maltose-binding protein (MBP) fusion proteins by cloning into pMAL-c2 (New England Biolabs Inc.). The entire rabbit NHE3 carboxyl-terminal construct extends from aa 475 to 832 (C3-832), and NHE3 C-terminal truncation constructs terminating at aa 711 (C3-711), 660 (C3-660), and 585 (C3-585) were generated by PCR and cloned into pET30a. The C-terminal 88 aa of NHE3 (C3-ter88) were generated by excising the NHE3 cDNA with StyI, followed by bluntending with Klenow fragment, digestion with XhoI, and subcloning into pET30b linearized with EcoRV and SalI.
Immunoprecipitation-Polyclonal antibody Ab2570 was raised in rabbit against the carboxyl-terminal 106 aa of E3KARP expressed as His 6 -tagged protein. Ab2570 was purified by (NH 4 ) 2 SO 4 precipitation, followed by affinity gel purification through Affi-gel Blue (Bio-Rad).
E3KARP was initially cloned into pET30a plasmid. The entire construct of the N-terminal His 6 and S-protein tags fused in frame with E3KARP was subcloned into pMT3 vector (Genetics Institute), resulting in pMT-e3karpHS. PS120/NHE3V fibroblasts were then stably transfected with pMT-e3karpHS, resulting in PS120/E3KARP-HS. To coexpress E3KARP and NHE3V, PS120/E3KARP-HS was transfected with pECE:NHE3V, and expression of NHE3V was selected by acidloading. NHE3V is rabbit NHE3 tagged at its carboxyl terminus with a VSVG epitope, as previously described (5). Transfected PS120 fibroblasts were lysed in lysis buffer (50 mM Tris-Cl, pH 7.4, 150 mM NaCl, 0.1 mM phenylmethylsulfonyl fluoride, 5 g/ml aprotinin, 1 M pepstatin, 1 mM iodoacetamide, and 1% Triton X-100), followed by centrifugation at 100,000 ϫ g at 4°C for 1 h. The lysate was incubated overnight with Ab2570. Immunocomplexes were separated by binding to protein A-Sepharose resin and were washed three times with lysis buffer prior to electrophoresis. The presence of immunoprecipitated recombinant E3KARP-HS was detected with S-protein-horseradish peroxidase conjugate as recommended by Novagen. The presence of co-immunoprecipitated NHE3V was detected with monoclonal anti-VSVG antibody mP5D4 (14).
Gel Overlay Assay-The entire NHE3 C terminus and the truncated mutants described above were labeled with [ 35 S]methionine using the TNT in vitro transcription-translation system (Promega). Prior to gel overlay assay, the relative yields of the in vitro translation products were estimated by resolving the labeled products on SDS-polyacrylamide gel, followed by autoradiography.
Domains of E3KARP were generated as described above. Three g of each recombinant protein was resolved by SDS-polyacrylamide gel electrophoresis and transferred onto nitrocellulose membranes. Nitrocellulose membranes were blocked by incubation for 1 h in 5% nonfat dried milk, 25 mM Tris-HCl, pH 7.4, and 150 mM NaCl and then were incubated with comparable amounts (typically 3-5 l/ml) of the 35 S-labeled entire NHE3 C terminus or truncated NHE3 C termini in 5% nonfat dried milk, 25 mM Tris-HCl, pH 7.4, 150 mM NaCl, and 0.2% Triton X-100 overnight at 4°C. Nitrocellulose membranes were washed four times with 25 mM Tris-HCl, pH 7.4, 150 mM NaCl, and 0.2% Triton X-100, and bound 35 S-labeled NHE3 C terminus was detected by autoradiography.
The N-terminal 302 aa of human ezrin (N-ezrin) were generated by PCR, and the PCR product was subcloned into pET30a. The N-ezrin was labeled with 35 S by in vitro transcription-translation and was used in gel overlay assay as described above.
Pull-down Assay-PS120/NHE3V fibroblasts were lysed in lysis buffer with 500 mM NaCl. The lysate was centrifuged at 100,000 ϫ g at 4°C for 1 h, and the supernatant was incubated with 10 g of recombinant E3KARP or various domains of E3KARP immobilized on Ni 2ϩ -NTA beads overnight at 4°C. The beads were washed four times with lysis buffer and twice with lysis buffer without Triton X-100, followed by boiling in SDS sample buffer. The samples were resolved by SDSpolyacrylamide gel electrophoresis and were immunoblotted with the mP5D4 antibody.
Immunofluorescence-PS120 fibroblasts were grown on tissue culture coverslips. Cells were serum-starved overnight, and when necessary, cells were treated with 0.1 mM 8-bromoadenosine 3Ј:5Ј-monophosphate (8-Br-cAMP) for 20 min before washing with PBS. After washing with PBS, cells were fixed at room temperature for 10 min with 3% paraformaldehyde in PBS, followed by permeabilization with 0.2% Triton X-100 in PBS for 10 min at room temperature. Cells were blocked in 15% goat serum, PBS, and 1% bovine serum albumin for 1 h at room temperature. Primary antibodies were incubated for 1 h at room temperature in 10% goat serum, PBS, and 1% bovine serum albumin at the following dilutions: 1:300 for polyclonal antibody Ab2570 (anti-E3KARP antibody), 1:50 for monoclonal antibody mP5D4 (anti-VSVG antibody), and 1:500 for monoclonal antibody 3C12 (anti-ezrin antibody; Sigma). Cells were then washed three times with PBS, bovine serum albumin, and 0.1% Tween 20 and incubated for 1 h with fluorochrome-conjugated secondary antibodies. Cells were washed three times with PBS, bovine serum albumin, and 0.1% Tween 20 and mounted with Prolong Antifade (Molecular Probes, Inc.). Nuclei were stained with Hoechst 33342 (Molecular Probes, Inc.). Cells were examined with an inverted microscope (Zeiss Axiovert) coupled to a confocal laser scanning unit (Bio-Rad MRC-600).

RESULTS
In Vivo Interaction between NHE3 and the Regulatory Factors-To demonstrate that E3KARP and NHE3 interact in vivo, we stably expressed E3KARP as a His 6 -and S-proteintagged fusion protein, E3KARP-HS, in PS120/NHE3V fibroblasts. To ensure that the tagging of E3KARP protein with His 6 and S-protein did not alter its function in cAMP-dependent inhibition of NHE3, we studied the activity of NHE3V in response to 100 M 8-Br-cAMP. Fig. 2 shows that the sodium-dependent alkalinization due to NHE3V activity is significantly inhibited in the presence of 8-Br-cAMP in PS120/NHE3V/ E3KARP-HS cells. The mean intracellular pH value of PS120/ NHE3V/E3KARP-HS in the absence or presence of 8-Br-cAMP was 7.38 Ϯ 0.036 and 7.28 Ϯ 0.027, respectively (p Ͻ 0.05 by unpaired t test; n ϭ 5).
We then determined if there was any interaction between NHE3V and E3KARP-HS in vivo. The antibody against the C-terminal 106 aa of E3KARP, Ab2570, recognized E3KARP-HS expressed in PS120/NHE3V fibroblasts (Fig. 3A). The C-terminal 106 aa of E3KARP used in preparation of Ab2570 show limited homology to the similar C-terminal domain of NHERF (5). Therefore, Ab2570 exhibited minimal cross-reactivity with NHERF (Fig. 3A). E3KARP-HS was immunoprecipitated with Ab2570 against E3KARP, and the success of immunoprecipitation was detected with S-proteinhorseradish peroxidase conjugate (Fig. 3B). Coprecipitated NHE3V was then detected with mouse monoclonal antibody mP5D4 against the VSVG epitope tag. As shown in Fig. 3B, NHE3V specifically co-immunoprecipitated with E3KARP-HS, demonstrating the interaction in vivo between NHE3 and E3KARP, consistent with our earlier finding (5).
E3KARP Binding Requires Both the Second PDZ Domain and the C Terminus-To determine which parts of E3KARP interact with the NHE3 C terminus, we expressed and purified the entire E3KARP and various subdomains of E3KARP as His 6 -tagged recombinant proteins. The entire NHE3 C terminus (C3-832) labeled with [ 35 S]methionine by in vitro transcription-translation was used as a probe. As shown in Fig. 4A, C3-832 bound to the full-length E3KARP protein, but not to PDZ domains expressed individually (P1 or P2) or as one unit (P1-2). We occasionally observed weak binding of C3-832 to the P1 and P1-2 domains, but this interaction did not occur consistently. The C3-832 probe did not bind to the C-terminal domain of E3KARP, but when the P2 and C-terminal domains were expressed as one unit (P2-C), C3-832 bound to the P2-C domain consistently. C3-832 also bound to the C42 construct, consistent with the interaction of clone C42 with NHE3C in the yeast two-hybrid system (5).
To confirm the above findings, we performed a "pull-down assay" on lysate prepared from PS120/NHE3V fibroblasts using bacterial recombinant E3KARP and its subdomains immobilized on Ni 2ϩ -NTA beads. The presence of NHE3V bound to the recombinant proteins was analyzed by immunoblotting using the mP5D4 antibody against the VSVG tag. As shown in Fig. 4B, NHE3V bound specifically to E3KARP and the P2-C and C42 constructs, but did not bind to the P1, P2, and Cterminal domains or to Ni 2ϩ -NTA itself.
To test the specificity of the binding of C3-832 to the P2-C domain, we expressed the domains of E3KARP as MBP fusion FIG. 2. Inhibition of NHE3 activity by 8-Br-cAMP in PS120/ NHE3V/E3KARP-HS fibroblasts. Serum-starved PS120/NHE3V/ E3KARP-HS fibroblasts were acidified with a NH 4 Cl prepulse, followed by perfusion with TMA buffer (15). Cells were then recovered either in Na ϩ buffer (broken line) or in Na ϩ buffer containing 0.1 mM 8-Br-cAMP (solid line). At the end of each experiment, the fluorescence ratio was calibrated using high potassium/nigericin buffer titrated at pH 6.1 and 7.2. Representative traces from five sets of experiments are shown.

FIG. 3. Interaction between E3KARP and NHE3 in vivo. A,
Western immunoblot using Ab2570. Thirty g of each lysate prepared from PS120/NHE3V, PS120/NHE3V/E3KARP-HS, and PS120/NHE3V/ NHERF was immunoblotted with rabbit polyclonal antibody Ab2570 against the C-terminal 106 aa of E3KARP. B, co-immunoprecipitation of NHE3V and E3KARP. Lysates prepared from PS120/NHE3V/ E3KARP-HS or PS120/NHE3V cells were incubated with the Ab2570 antibody raised against the C-terminal 106 aa of E3KARP. In the upper panel, immunoprecipitates (IP) were immunoblotted with the mP5D4 antibody against the VSVG tag. The right lane shows the presence of NHE3V in the lysate used for the immunoprecipitation. In the lower panel, immunoprecipitated E3KARP-HS was detected using S-proteinhorseradish peroxidase (HRP) conjugate only in PS120/NHE3V/ E3KARP-HS cells incubated with the Ab2570 antibody, demonstrating the specificity of immunoprecipitation. The expression of E3KARP-HS in PS120/NHE3V/E3KARP-HS cells is shown in the right lane. Molecular mass standards (in kilodaltons) are indicated on the right.

FIG. 4. Interaction with E3KARP requires the second PDZ domain and the carboxyl-terminal tail of E3KARP.
A, domains of E3KARP were expressed as His 6 -tagged fusion proteins and purified. Three g of each sample was run on SDS-polyacrylamide gels and transferred onto nitrocellulose membrane. The membrane was overlaid with 35 S-labeled NHE3C as described under "Experimental Procedures." Molecular mass standards (in kilodaltons) are indicated on the right. B, lysate prepared from PS120/NHE3V fibroblasts was incubated with His 6 -tagged E3KARP or domains of E3KARP immobilized on Ni 2ϩ -NTA beads or with Ni 2ϩ -NTA beads alone, and bound proteins were resolved by SDS-polyacrylamide gel electrophoresis. Bound NHE3V was identified with the mP5D4 antibody.
proteins. These were then overlaid with 35 S-labeled NHE3C. As shown in Fig. 5A, C3-832 bound to MBP/E3KARP and MBP/P2-C, but not MBP/P1, MBP/P2, MBP/C, or MBP, confirming the results shown in Fig. 3A. Subsequently the specificity of interaction was tested by incubating the blots with labeled C3-832 and competing peptides expressed as His 6tagged proteins (20 g/ml) as inhibitors. Fig. 5B shows that in the presence of E3KARP-His 6 protein, 35 S-labeled NHE3C binding was significantly reduced. Similarly, the P2-C domain effectively blocked NHE3C binding to E3KARP and the P2-C domain. In contrast, the P1 domain showed no effect on the binding, consistent with the weak binding observed in the gel overlays.
E3KARP Binds to an Internal Domain of the NHE3 C Terminus-We next determined which part of the NHE3 C terminus interacts with E3KARP. Gel overlay assays were performed on E3KARP using the 35 S-labeled truncated NHE3 C-terminal constructs. As shown in Fig. 6, the entire NHE3 C terminus (C3-832) and C3-711, which ranges from aa 475 to 711, bound to E3KARP. Similarly, C3-660, corresponding to aa 475-660, bound to E3KARP, although the binding of C3-660 to E3KARP was always significantly lower than that of C3-711. However, C3-585 did not interact with E3KARP. Because many PDZ domains specifically interact with the carboxyl termini of protein targets (12,16), the carboxyl-terminal 88 aa of NHE3 (C3-ter88) were expressed and used as a probe for a gel overlay to determine whether this part of NHE3 contributed to the interaction with E3KARP. However, C3-ter88 did not interact with E3KARP. This lack of binding of the C-terminal 88 aa of NHE3 to the regulatory proteins is consistent with the fact that E3KARP interacts with NHE3V even though the C terminus of NHE3 is blocked by a VSVG epitope (Fig. 3B).
Cellular Distribution-We determined the distribution of E3KARP and NHE3V exogenously expressed in PS120 fibroblasts. We first determined the distribution of E3KARP in PS120 fibroblasts in the absence of NHE3. PS120 fibroblasts were transfected with pMT-e3karpHS and selected by 600 g/ml hygromycin. Cells were serum-starved overnight and treated with 100 M 8-Br-cAMP for 15 min prior to fixation. Fig. 7A shows that E3KARP is diffusely distributed throughout the cytosol, with some staining at or near the plasma membrane. Treatment with 8-Br-cAMP did not affect the distribution of E3KARP. In contrast to a previous report that SIP-1, a probable splice variant of E3KARP, was exclusively present in nuclei (17), E3KARP was not present in nuclei since the labeling of E3KARP did not overlap with that of nuclei, which were stained with Hoechst 33342.
E3KARP distribution was also studied in the presence of NHE3V. Fig. 7B reveals that NHE3V showed predominant staining in juxtanuclear regions, consistent with an earlier report suggesting that NHE3 is located in recycling endocytic vesicles (18). In addition to the juxtanuclear staining, localization of NHE3V at or near the plasma membrane was seen. In contrast, E3KARP again was predominantly present in the cytosol, with distinct staining at the plasma membrane as in Fig. 7A. With NHE3V coexpressed, we often observed more membrane staining of E3KARP than in the absence of NHE3V (Fig. 7, compare A (panel a) and B (panels b and e)). However, this requires further studies that enable quantification of surface membrane-associated E3KARP. Unlike NHE3V, predominant staining at the perinuclear region was not observed for E3KARP. Double staining showed that NHE3 and E3KARP did not overlap throughout the cells, but co-localized at the plasma membrane as well as some areas within the cytosol, but not in the juxtanuclear regions. When treated with 8-Br-cAMP, the co-localization of NHE3V and E3KARP did not differ significantly compared with the basal condition. Thus, neither NHE3V nor E3KARP appears to move in response to cAMP.
The C Terminus of E3KARP Interacts with the N Terminus of Ezrin-Recent studies by Reczek et al. (19) and Murthy et al. (20) have shown that the human homolog of NHERF, EBP50, binds to the ezrin-radixin-moesin family. Since both E3KARP and NHERF interact with NHE3 and seemingly have the same function in the cAMP-dependent inhibition of NHE3, we determined if ezrin can also interact with E3KARP. The N-ezrin was labeled with [ 35 S]methionine by in vitro transcription-translation and was used as a probe for gel overlay. NHERF (5, 6) expressed as a His 6 -tagged fusion protein was used as a positive control for the N-ezrin binding. Fig. 8A shows that the N-ezrin bound to MBP/E3KARP, but not MBP itself, indicating specific interaction between the N terminus of ezrin and E3KARP. The interaction of the N-ezrin with E3KARP was as strong as that with NHERF. To determine which part of E3KARP binds to the N-ezrin, interaction between the N-ezrin and the subdomains of E3KARP expressed as His 6 -tagged fusion proteins was studied by gel overlay assay. As shown in Fig.  8B, the N-ezrin did not bind to the individual PDZ domains (P1 and P2) or to P1-2 as one unit. However, the N-ezrin bound to the C-terminal and P2-C domains of E3KARP. Interestingly, FIG. 5. Interaction between E3KARP and the NHE3 C terminus can be inhibited by specific peptides. A, E3KARP and various domains of E3KARP were expressed as MBP fusion proteins and overlaid with the 35 S-labeled entire NHE3 cytoplasmic domain (C3-832) as described under "Experimental Procedures." B, specificity of the interaction was examined by incubating blots in the presence of competing peptides. Overlay assay with 35 S-labeled NHE3C was done in the presence of 15 g/ml His 6 -tagged E3KARP, P2-C domain, or P1 domain.
FIG. 6. E3KARP binds to an internal sequence within the cytoplasmic domain of NHE3. Recombinant E3KARP was overlaid with comparable amounts of 35 S-labeled NHE3 C-terminal constructs (C3-832, C3-711, C3-660, C3-585, and C3-ter88) as described under "Experimental Procedures." The relative amounts of the NHE3 C-terminal constructs were estimated prior to the overlay assay by resolving and quantitating the relative yields by autoradiography. the N-ezrin did not interact with the C42 construct, which is essentially P2-C lacking the carboxyl-terminal 23 aa residues. The lack of N-ezrin binding to the C42 construct suggests that the ezrin-binding motif is located within the carboxyl-terminal 23 aa residues of E3KARP (Fig. 8C). Fig. 8D shows the distribution of E3KARP and ezrin in PS120/NHE3V/E3KARP-HS cells. Although both proteins were largely expressed in the cytoplasm as expected, some membrane staining was also evident. Double staining of endogenous ezrin and exogenous E3KARP revealed that ezrin and E3KARP are co-localized in some areas along the plasma membrane. Treatment with 8-Br-cAMP for 20 min did not alter the distribution of E3KARP or ezrin or the co-localization of these two proteins. DISCUSSION In this study, we determined the interaction between NHE3 and E3KARP. We specifically wanted to address whether the PDZ domains in E3KARP contribute to the interaction with NHE3. This was not only because PDZ domains were recently shown to be the basis for protein-protein interaction, but also because one of the clones obtained in our initial two-hybrid screen encompassed the second PDZ domain and a large part of FIG. 7. Cellular distribution of E3KARP and NHE3 in PS10 fibroblasts. A, subcellular distribution of E3KARP in PS120/E3KARP-HS fibroblasts. PS120/E3KARP-HS cells were serum-starved overnight, and immunofluorescence staining was carried out under basal conditions (panel a) or after incubation with 100 M 8-Br-cAMP for 15 min (panel b). E3KARP was stained by indirect immunofluorescence using Ab2570, followed by anti-rabbit antibody conjugated with fluorescein isothiocyanate (red), and analyzed by confocal microscopy using a ϫ40 lens. Nuclei were stained with Hoechst 33342 (green). B, co-localization of NHE3V and E3KARP in PS120/NHE3V/E3KARP-HS fibroblasts. NHE3V was stained with the mP5D4 antibody against the VSVG tag, followed by Cy5-labeled anti-mouse antibody (red) (panels a and d). E3KARP was stained using Ab2570, followed by anti-rabbit antibody conjugated with fluorescein isothiocyanate (green) (panels b and e). Immunofluorescence staining was performed on serum-starved cells under basal conditions (panels a-c) or after a 15-min incubation with 100 M 8-Br-cAMP (panels d-e). The yellow color in panels c and f shows co-localization of NHE3V and E3KARP on the plasma membrane. FIG. 8. Interaction of the N-terminal domain of ezrin with E3KARP. A, 3 g each of bacterially expressed MBP/E3KARP, MBP, and NHERF-His 6 was overlaid with the 35 S-labeled N-terminal 302 aa of human ezrin as described under "Experimental Procedures." B, His 6 -tagged E3KARP and domains of E3KARP were overlaid with the 35 S-labeled N-terminal 302 aa of human ezrin. C, the amino acid sequences of the C-terminal 30 aa of E3KARP and NHERF are aligned. Identical amino acid residues are indicated by vertical lines. D, shown is the co-localization of E3KARP and ezrin in PS120/NHE3V/E3KARP-HS fibroblasts. Immunofluorescence staining was performed on serum-starved cells under basal conditions (panels a-c) or after a 15-min incubation with 100 M 8-Br-cAMP (panels d-f ). E3KARP (panels a and d) was stained as described in the legend to Fig. 6. Ezrin (panels b and e) was stained with mouse monoclonal antibody 3C12, followed by Cy5-labeled anti-mouse antibody. Images were collected with ϫ40 objectives. E3KARP and ezrin are observed to co-localize on the plasma membrane (panels c and f). the C terminus of E3KARP (5). Based on this, we anticipated that NHE3 might interact with E3KARP through the second PDZ domain.
Studies on protein-protein interaction via PDZ domains showed that distinct PDZ domains interact with specific carboxyl termini with a consensus sequence of E(S/T)X(V/I) (11,12). However, certain PDZ domains can bind other PDZ domains. For example, the PDZ domain of nNOS can bind to the second PDZ domain of ␣-syntrophin in skeletal muscle (21). Interaction of E3KARP with NHE3 represents a new mode of binding for a PDZ domain. In this study, we found that the second PDZ domain was a major constituent in the interaction with the NHE3 C terminus, but the second PDZ domain alone was not sufficient for the interaction. Instead, the P2 domain and the C terminus of E3KARP formed a unit that interacted with NHE3. This is not the first case where a PDZ domain does not work in isolation. In other cases, paired PDZ domains were required for interaction (22,23). Band 4.1 binds to the first and second PDZ domains of hDlg, a human homolog of the Drosophila tumor suppressor, as a unit, but not to individual PDZ domains (22). Similarly, binding of syndecan, a major cellsurface heparan sulfate, to the PDZ domain-containing protein syntenin requires two PDZ domains of syntenin (23). However, in this study, the P1-2 domain and also the P1 domain of E3KARP displayed weak binding characteristics that were not observed consistently, and we thus dismissed this as a possible artifact of the in vitro conditions. The interaction of the NHE3 C terminus with E3KARP differs from previously defined cases in that it required the second PDZ domain and a non-PDZ region, the C-terminal portion. Why the second PDZ domain of E3KARP does not work in isolation is not clear.
The P2-C domain of E3KARP is shown to interact with an internal domain of NHE3 between aa 585 and 660 of the NHE3 C-terminal tail. Since the relative E3KARP binding to C3-660 was significantly lower than that to C3-711, it is conceivable that the domain between aa 660 and 711 may also take part in binding to E3KARP. Concerning the mapping of the E3KARPbinding domain within the NHE3 C terminus, our analysis was limited to the gel overlay assays since we were not able to express the rabbit NHE3 C-terminal constructs as recombinant proteins. We attempted to express them as MBP, His 6 -tagged, and glutathione S-transferase fusion proteins in E. coli, but these were all unsuccessful, resulting in either the formation of inclusion bodies or total absence of expression. Therefore, identification of the exact locations and thus amino acid residues of NHE3 that are involved in the interaction with E3KARP requires further investigation. Regardless, it is noteworthy that deletion of the domain between aa 579 and 684 obliterated the cAMP-dependent inhibition of NHE3 (13). Recently, Kurashima et al. (4) identified Ser-605 of rat NHE3 as the sole site of phosphorylation in NHE3 by PKA. However, mutation of Ser-605 (S605A) did not completely abolish the cAMP-dependent inhibition of NHE3 activity, but decreased the extent of the inhibition by ϳ50%. In fact, to completely abolish the inhibition of NHE3 by cAMP, simultaneous mutations of Ser-605 and Ser-634 were needed, although Ser-634 is not phosphorylated by PKA. The functional role of Ser-634 remains unknown, but it is enticing to think that Ser-634 may participate in the interaction with the regulatory proteins. Preliminary experiments indicate that NHERF also binds to the same region in NHE3 as E3KARP. 3 Of note, this is not the only case in which a PDZ domain interacts with an internal domain. For example, the second PDZ domain of the retinal protein Inad binds an internal motif near the C terminus of the transient receptor potential-activated Ca 2ϩ channel (24). PDZ domain-containing proteins often function as scaffold proteins, resulting in clustering of proteins that interact with PDZ domains. For example, Shaker subfamily K ϩ channels and the N-methyl-D-aspartate receptors in central neurons cluster at specific subcellular sites via interactions with the PDZ domains in the prominent postsynaptic density protein PSD-95 (11,12). Confocal immunofluorescence studies of PS120 fibroblasts revealed that E3KARP is largely distributed in the cytosol (Fig. 7). In addition, there was distinct membrane staining associated with E3KARP. How E3KARP (and NHERF) transduces the cAMP effect on NHE3 is not known. However, it may involve redistribution of E3KARP in response to phosphorylation by PKA. However, we found in this work that distribution of E3KARP was not changed in the presence of 8-Br-cAMP. We also observed that NHERF distribution, which is similar to E3KARP distribution, was not affected by 8-Br-cAMP. 3 The lack of redistribution of E3KARP (and NHERF) is perhaps consistent with the fact that the phosphorylation level of neither protein was changed in response to 8-Br-cAMP. 4 This study showed that NHE3V and E3KARP are co-localized at the plasma membrane, consistent with the interaction between NHE3V and E3KARP in vivo and the E3KARP-dependent regulation of NHE3 by cAMP. However, the distribution of NHE3V in PS120 fibroblasts is largely distinct from that of E3KARP, with NHE3V localized at perinuclear regions and on the surface membrane. The perinuclear staining is thought to represent NHE3 proteins present in recycling endosomes (18). Distribution of NHE3V was not changed by the presence of 8-Br-cAMP or coexpression of E3KARP. These data show that the inhibition of NHE3 in response to an elevated cAMP level may not involve redistribution of either NHE3 or E3KARP within the resolution of confocal microscopy. However, we cannot rule out a small change in E3KARP or NHE3 distribution undetectable by confocal microscopy, and this requires other quantitative analyses.
Since PDZ domain-containing proteins function as scaffold proteins, we have recently been contemplating a model in which E3KARP and/or NHERF functions as a scaffold. We examined if E3KARP and NHERF might be A kinase anchoring proteins (AKAPs) (25), but found that these regulatory proteins are not (26). Recently, NHERF/EBP50 was shown to bind to ezrin and other ezrin-radixin-moesin family proteins (19,20). Since ezrin has been shown to directly bind PKA (27), the binding of NHERF to ezrin suggested that the NHERFezrin complex may be a linker between NHE3 and PKA (19,20). E3KARP and NHERF share a significant homology, and therefore, we determined if E3KARP is an ezrin-binding protein. Gel overlay study showed that the N-terminal 302 aa of human ezrin bind to both NHERF and E3KARP with comparable affinity. We found that the interaction with ezrin occurs through the C-terminal portion of E3KARP. Our current study further localized the ezrin-binding motif to the C-terminal 23 aa residues (Fig. 8C). Similarly, a recent report showed that ezrin also binds to the carboxyl-terminal 30 aa of human EBP50 (28). A search for proteins with a similar motif did not identify proteins other than NHERF and its homologs, suggesting that this domain of E3KARP may represent a unique ezrinbinding motif. The mechanistic nature of the PKA-dependent inhibition of NHE3 is yet to be determined. The current interpretation of data indicates that the increase in phosphorylation of NHE3 is essential for this inhibition (4). Concurrently, the presence of a regulatory protein, either E3KARP or NHERF, is also necessary. Since ezrin is an AKAP that binds to the reg-ulatory subunit of type II PKA (27), the binding of ezrin to NHERF suggested a model in which NHERF may scaffold NHE3 and ezrin, which in turn recruit PKA into the vicinity of NHE3, allowing PKA to phosphorylate NHE3. Whether the phosphorylation of NHE3 by PKA and the presence of the regulatory proteins are dependent processes needs further investigation.